Debian Bug report logs - #921004
AMD Radeon RX 580: no GL display, "amdgpu: The CS has been cancelled because the context is lost"

Display info messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Jean-Dominique Frattini <jd.frattini@free.fr>

To: submit@bugs.debian.org

Subject: amdgpu: The CS has been cancelled because the context is lost.

Date: Thu, 31 Jan 2019 14:40:41 +0100

Package: xserver-xorg-video-amdgpu

Version: 18.1.0-1

Severity: critical

Debian: Stretch amd64

Regression: Yes

Graphic card: AMD Radeon Rx 580


Good Morning,

since the latest update of xserver-xorg-video-amdgpu and 
firmware-amd-graphics, most GL applications do not display anything and 
display this error message in the console:

amdgpu: The CS has been cancelled because the context is lost.

This makes Debian unusable for developing and running GL programs.

-- 
Jean-Dominique Frattini

Message #10 received at 921004@bugs.debian.org (full text, mbox, reply):

From: Jean-Dominique Frattini <jd.frattini@free.fr>

To: 921004@bugs.debian.org

Subject: downgrade to firmware-amd-graphics_20180825-1_all.deb

Date: Thu, 31 Jan 2019 15:07:45 +0100

Good Morning,

downgrading to firmware-amd-graphics_20180825-1_all.deb seems to fix the 
issue.

Please let me know if I should report this bug directly to 
firmware-amd-graphics.

Message #15 received at 921004@bugs.debian.org (full text, mbox, reply):

From: Bernhard Übelacker <bernhardu@mailbox.org>

To: Jean-Dominique Frattini <jd.frattini@free.fr>

Cc: 921004@bugs.debian.org, 921114@bugs.debian.org, 921145@bugs.debian.org

Subject: Re: amdgpu: The CS has been cancelled because the context is lost.

Date: Sat, 2 Feb 2019 17:28:39 +0100

Hello Jean-Dominique Frattini,
I am not involved in packaging any of these packages,
but I noted that you opened the last three days nearly
the same issue against three different packages?
What are you trying to achive that way?
And you noted versions contained in current Buster,
but you claim it is Stretch?
You write it is a regression - do you know the package
versions or the date when it was working like expected?
Also giving an example application that triggers that
error message could maybe make the bug report more
attractive to get some one having the matching hardware
looking at it.

Kind regards,
Bernhard

Message #20 received at 921004@bugs.debian.org (full text, mbox, reply):

From: Jean-Dominique Frattini <jd.frattini@free.fr>

To: Bernhard Übelacker <bernhardu@mailbox.org>

Cc: 921004@bugs.debian.org, 921114@bugs.debian.org, 921145@bugs.debian.org

Subject: Re: amdgpu: The CS has been cancelled because the context is lost.

Date: Sat, 2 Feb 2019 19:02:06 +0100

Sorry if I created a burden. I didn't wanted to achieve anything 
particular except to find the good location where to report the bug. In 
fact I am not really sure which package is 'responsible' of this bug: At 
first glance, this is a driver issue because the bug is not reproducible 
with GPUs from other manufacturers (tested with nVidia and Intel). 
Downgrading the firmware resolves partially the issue, therefore the 
firmware package maintainers might have clue about this. Finally, the 
error message is raised by Mesa.

For about the Debian version, this is a mistake on my side. It is indeed 
in Buster.

It is a regression because prior to upgrading to the latest {firwmare, 
xserver-xorg, Mesa} there was no bugs.

There were no bugs with the previous version of firmware-amd-graphics, 
that is 20180825-1 along with the previous version of Mesa, that is the 
18.2 branch.

Regards.

Le 02/02/2019 à 17:28, Bernhard Übelacker a écrit :
> Hello Jean-Dominique Frattini,
> I am not involved in packaging any of these packages,
> but I noted that you opened the last three days nearly
> the same issue against three different packages?
> What are you trying to achive that way?
> And you noted versions contained in current Buster,
> but you claim it is Stretch?
> You write it is a regression - do you know the package
> versions or the date when it was working like expected?
> Also giving an example application that triggers that
> error message could maybe make the bug report more
> attractive to get some one having the matching hardware
> looking at it.
>
> Kind regards,
> Bernhard

Message #25 received at 921004@bugs.debian.org (full text, mbox, reply):

From: Simon McVittie <smcv@debian.org>

To: Jean-Dominique Frattini <jd.frattini@free.fr>, 921114@bugs.debian.org

Cc: at46 <at46n@t-online.de>, Ben Hutchings <ben@decadent.org.uk>, Gijs Schröder <gijsschroder@gmail.com>, Bernhard Übelacker <bernhardu@mailbox.org>, xserver-xorg-video-amdgpu@packages.debian.org, libglx-mesa0@packages.debian.org

Subject: Re: Bug#921114: AMD Radeon RX 580: no GL display, "amdgpu: The CS has been cancelled because the context is lost"

Date: Thu, 14 Mar 2019 09:07:02 +0000

Control: reassign 921004 firmware-amd-graphics 20190114-1
Control: reassign 921145 firmware-amd-graphics 20190114-1
Control: forcemerge 921114 921004 921145
Control: retitle 921114 AMD Radeon RX 580: no GL display, "amdgpu: The CS has been cancelled because the context is lost"
Control: tags 921114 + moreinfo
Control: affects 921114 + libglx-mesa0 xserver-xorg-video-amdgpu

On Fri, 01 Feb 2019 at 18:16:41 +0100, Jean-Dominique Frattini wrote:
> since the latest update of xserver-xorg-video-amdgpu and
> firmware-amd-graphics [in buster], most GL applications do not display
> anything and display this error message in the console:
> 
> amdgpu: The CS has been cancelled because the context is lost.

You say "most GL applications". Is glxgears (from the mesa-utils package)
affected? If not, please give some examples of specific GL applications
that are affected (smaller/simpler applications are better examples
here than larger/more complicated ones, so a simple GL game would be a
good test).

If you report more bugs in future, please report them with the reportbug
tool if possible. It provides information that package maintainers
will need if they are going to be able to solve bugs like this, like
the exact versions of all the packages involved (including the kernel)
and the relevant system log entries.

Some commands that might provide useful information here:

    reportbug --template libglx-mesa0
    reportbug --template firmware-amd-graphics
    reportbug --template xserver-xorg-video-amdgpu

If possible, please get the system into a state where you see the bug
again, then run those reportbug commands and sent their output to the
bug's email address. If that isn't possible, please be as clear and
specific as possible about what was different when the bug was present
(for instance the exact versions of the relevant packages).

In the other bugs you opened at the same time, you mentioned this
being a regression in an upgrade of firmware-amd-graphics, Mesa and
xserver-xorg-video-amdgpu, all together. I've merged the three separate
bugs you opened, and reassigned the merged bug to firmware-amd-graphics
for now. Debian contributors can reassign it to whatever package is
most appropriate if it turns out to be a bug in a different package
than the one you first thought: there is no need to open separate bugs
in multiple places.

> Note that downgrading to firmware-amd-graphics_20180825-1_all.deb seems to
> fix the issue.

Please confirm: if you downgrade firmware-amd-graphics to version
20180825-1, leave all other packages at their latest versions in
buster, and reboot, does that resolve this problem? If you upgrade
firmware-amd-graphics to version 20190114-1 and reboot again, does the
problem come back?

On Wed, 06 Mar 2019 at 21:43:01 +0100, Gijs Schröder wrote:
> If this is of any help, I'm on Buster with an AMD RV710 for a graphics card.
> Both glxinfo and glxgears run without apparent issues after booting with
> firmware-amd-graphics version 20190114-1 installed.

This bug is probably specific to one hardware generation. If I'm reading
Wikipedia correctly, the RX 580 is from a 2017 generation codenamed
"Arctic Islands" and based on Polaris GPUs, whereas the RV 710 is much
older (from a 2008 generation identified as the R700 or Radeon HD 4000
series).

firmware-amd-graphics 20180825-1 does seem to contain updates to the
firmware for (at least some) Polaris GPUs.

    smcv

Message #52 received at 921004@bugs.debian.org (full text, mbox, reply):

From: maximilian attems <maks@stro.at>

To: Jean-Dominique Frattini <jd.frattini@free.fr>, 921004@bugs.debian.org

Subject: Re: AMD Radeon RX 580: no GL display, "amdgpu: The CS has been cancelled because the context is lost"

Date: Tue, 23 Mar 2021 19:04:33 +0100

tags 921004 moreinfo
stop

can you still reproduce this trouble with uptodate Debian 11 Testing?


thank you for your report and sorry for the late reply.

Message #57 received at 921004-done@bugs.debian.org (full text, mbox, reply):

From: maximilian attems <maks@stro.at>

To: 921004-done@bugs.debian.org

Subject: Re: AMD Radeon RX 580: no GL display, "amdgpu: The CS has been cancelled because the context is lost"

Date: Wed, 24 Mar 2021 11:41:02 +0100

Version: 20210208-4

Closing as submitter no longer reachable and bug assumed fixed in
up to date Debian testing.

Send a report that this bug log contains spam.